×

Image Segmentation Using Hierarchical Unsupervised Segmentation and Hierarchical Classifiers

  • US 20140037198A1
  • Filed: 08/06/2012
  • Published: 02/06/2014
  • Est. Priority Date: 08/06/2012
  • Status: Active Grant
First Claim
Patent Images

1. An image segmentation method comprising:

  • generating a hierarchy of regions by unsupervised segmentation of an input image;

    describing each region of the hierarchy with a respective region feature vector that is representative of the hierarchy region;

    identifying hierarchical structures in the hierarchy, each of the hierarchical structures comprising a parent hierarchy region and its respective child hierarchy regions;

    describing each of the input image hierarchical structures with a respective hierarchical feature vector, the hierarchical feature vector being based on the region feature vectors of the respective parent hierarchy region and child hierarchy regions;

    with a hierarchical classifier component, classifying each of the input image hierarchical structures according to a set of predefined classes, the hierarchical classifier component having been trained with hierarchical feature vectors of hierarchical structures of training images, the training images comprising semantic regions labeled according to the set of predefined classes; and

    segmenting the input image into a plurality of semantic regions based on the classification of the hierarchical structures.

View all claims
  • 1 Assignment
Timeline View
Assignment View
    ×
    ×